We evaluated Code Llama against existing solutions on both HumanEval & MBPP.
– It performed better than open-source, code-specific LLMs & Llama 2.
– Code Llama 34B scored the highest vs other SOTA open solutions on MBPP — on par w/ ChatGPT. More info https://
bit.ly/45JiPwJ
CODE
-
Code Llama Outperforms Open-Source Solutions on HumanEval MBPP
By
–
-
LangChain releases chat loaders for Llama fine-tuning event
By
–
If you are in SF tmrw… there is a great event happening tmrw around finetuning Llama In anticipation of that we released some new loaders to help load chat data into an easy-to-use format: https://
blog.langchain.dev/chat-loaders-f
inetune-a-chatmodel-in-your-voice/
… Join the event here: -
RAG Pipeline Technical Overview and Implementation Guide
By
–
A great overview of what's going on under the hood in a RAG pipeline Thanks for writing this @czue
! -
Python ecosystem shifting to Rust over C++ for performance
By
–
Or more like a competitor to Rust since the Python ecosystem is moving to that instead of C++ (eg see polars and some others)? And for GPU & AI stuff a competitor to Triton/CUDA?
-
Code Llama Now Available in Hugging Face Playground
By
–
You can try Code Llama now in the Code Llama playground @huggingface space — it's also available in the Hugging Face ecosystem, starting with transformers version 4.33.
-
ML Engineering: Inventing New Design Patterns for Modern AI
By
–
One nice thing about machine learning engineering is that, because the field is so new, you get to constantly invent new design patterns. Classic software engineering wisdom is still quite useful, but there's often great value in figuring out how to apply it in novel ways.
-
Potential Competitor to Triton for Python Users
By
–
Yeah, we’ll see. It’s probably not going to be used by the average Python users, but I can see it becoming a competitor for Triton though.
-
Multiple Embeddings Strategy for Enhanced Document Retrieval
By
–
One solution is to start creating not one but MULTIPLE embeddings per document This was the basic realization with our ParentDocumentRetriever ~2 weeks ago, but it's really much more general than that There are many ways to create multiple embeddings
-
Multi Vector Retriever: Advanced Embedding Strategies for AI
By
–
Multi Vector Retriever The basic idea: you store multiple embedding vectors per document. How do you generate these embeddings? Smaller chunks (this is ParentDocumentRetriever)
Summary of document
Hypothetical questions
Manually specified text snippets Quick -
HALLM Agents Learn from Mistakes Through Feedback Loops
By
–
#HALLM agent can learn from mistakes (there's a feedback loop) https://t.co/bBzuVJSgyW
— Marek Rosa | European🇪🇺 | South African🇿🇦 (@marek_rosa) 25 août 2023#HALLM agent can learn from mistakes (there's a feedback loop)